Server Deployments – Week 17
SLS Main Channel
On Tuesday 23rd April, the SLS Main channel received the update package deployed to the three Release Candidate channels in week 16, primarily comprising the new server-side LSL Animation Override capabilities, complete with a fix for BUG 2164, wherein the new capabilities could conflict with built-in animation poses in chairs, etc., as discussed in my week 15 updates. This deployment additionally includes the slight region performance improvement when there are no pathfinding characters present. Release notes are available.
Release Candidate Channels
On Wednesday 24th April, the RC channels should receive the following updates:
- BlueSteel and LeTigre: should gain a new project which brings preliminary server-side support for experience permissions – release notes
- Magnum: should gain a new server maintenance project. This update brings some new minor features to LSL, and fixes some crash modes – release notes
So the long-awaited experience tools / advanced content creation tools permissioning system looks to be finally on its way, a little under a year since an exploit caused them to be withdrawn after their initial deployment.
Development Viewer Updates
The SL development viewer has been through a rapid series of updates over the last few days of week 16 and the start of week 17, with version 22.214.171.12444 released on Monday April 22nd. While release notes are not available, it is thought it includes (among other things), the fix for music stuttering every few seconds when using FMOD EX, as submitted to LL by Latif Khalifa (see OPEN-173).
Particle Project Viewer Soon (?)
It appears a new particle project viewer may be on the horizon in the near future. If it does appear, it is likely to be primarily aimed at testing a new capability to help deal with griefing attacks which use particle emitters. Speaking at the Simulator User Group meeting, Simon Linden said:
We’re doing some testing and may get a project viewer out, which would allow you to test it and (I believe) let other viewers check out the source code. This is right-click on a particle, and it kills the generator … We’ve tossed around a few ideas about blocking particles in other ways but definitely want to get these first steps out the door.
This sparked further discussion on ways and means to help stop particle griefing, which has been on the increase across the Mainland of late. However, given that particles are a viewer-side effect, “stopping” them is actually easier said than done from a server-side perspective. One idea which gained some interest, if it is feasible, would be a block of any particle emitters belonging to an avatar banned from a region / parcel.
In the meantime there are means of stopping the viewer from rendering all particles – such as CTRL – Alt – Shift – +, or going to Preferences > Graphics > and dropping the Particle Count slider to zero. These again only work in your own world view, and while not always ideal, do present an option.
Asked if such a particle project viewer might include the new LSL particle capabilities already deployed server-side, Simon could only say, “Hopefully yes … I’ve been working on getting that chunk of code solid for the last few days. Unfortunately it’s pretty easy to crash at this point and so it’s not ready for consumption.”
Server-side Baking / Appearance
The viewer-side code for SSB/A is starting to appear in more TPVs:
- Cool Viewer has had SSB/A support in the “Experimental” branch for a while, and this moved to the “Stable” and “Legacy” versions on April 20th – see the Cool VL website and release notes
- Singularity released version 1.8.0 with SSB/A support on April 21st, which I’ve reviewed
- Firestorm released version 4.4.0 with SSB/A support on April 22nd, which I’ve reviewed
- Kukua has had an experimental / development version since the start of April (version 3.5.1) which also incorporates SSB/A.
The status of the SL beta viewer with the SSB/A code will take place around the middle of week 17, when a decision taken as to whether to incorporate the code into the SL release viewer or run a further beta release. The expectation is that the code will move to the release channel unless a last-minute significant issue is found within the code which prompts a further run in beta.
Assuming the code does make it to the release channel, it is likely that a release viewer will appear early in week 18 rather than late week 17, due to the time required to test and QA a build.
SUN-72 – Fix Submitted
One issue which is known to exist within the current beta release of the SSB/A viewer code and which has caused considerable problems is SUN-72 – if you have inventory items with special characters (including the likes of asterisks) in their names, they will fail to load. Similarly the use of accented characters (e.g. such as René) in things like chat log path names, the logs are not saved.
Nicky Dasmijn of the Firestorm team developed a patch for this specific problem which is already incorporated into the recent Firestorm 4.4.0 release, and which has been contributed to, and accepted by, the Lab for integration into the SSB viewer code. Hopefully, it should be appearing in the next SSB viewer release, whether beta of SL release viewer.
Sun-38 – Avatar / z-offset
As noted in my last major update on SSB/A, the ability to offset an avatar’s height on-the-fly to accommodate various animations (e.g. kneeling), or to adjust an avatar’s height when sitting / walking of floors, etc., and which will effectively be “broken” as reported in SUN-38, remains an issue for many.
While the Lab have produced an alternative approach which addresses some issues around avatar height offset using new appearance sliders, and which Nyx Linden and others have been continuing to tweak, there is no solution on the horizon for maintaining any form of on-the-fly adjustment; nor does the proposed solution work with no-mod shapes, as noted when I first reported on the “fix”.
During the Content Creation User Group meeting on Monday, April 22nd, a request was made for the Lab to not shut-down the existing baking service until such time as SUN-28 is “solved”, which prompted Nyx to comment:
SUN-38 is not currently considered a blocking issue for our initial release. It’s on our list to investigate, but we don’t have a patch or update immediately. There is a new hover parameter, which should work for attachments that affect your avatar, but the bug reported is also discussing the need to adjust the offset on the fly without changing wearables. Having both the new and the old system enabled is not an option … We are investigating options, but it is not a hard requirement for the initial release.
So, while the current situation is unlike to change prior to the SSB switch being thrown, again as per my last update, this doesn’t mean the issue is entirely closed from LL’s perspective.
A problem has been found in that some high-end audio chips using by OSX dislike having to initialise FMOD Ex at 44kHz (a requirement for FMOD Ex to function on Windows XP systems), which can cause audio corruption after the viewer process has ended. A JIRA on the matter was due to be raised on Monday 22nd April, in order to allow further investigation to take place in the hope that a fix can be provided prior to the code moving to the beta viewer.
Sovereign Engineer, who has reported the above issue, has also provided a script for packaging FMOD Ex (see OPEN-74), which may provide assistance to TPVs in using FMOD Ex in their builds.
Of Viewers and Versions
In my last update, I commented on the problem of communications and the number of different versions of the various viewers in circulation, when it comes to ensuring everyone about the forthcoming SSB changes. During a general discussion about viewers at the Open-source Dev meeting, Oz Linden indirectly underlined this with some interesting snippets of data based on LL’s most recent viewer stats:
- Firestorm is the most popular viewer in terms of number of user minutes – and there are some 329 different versions being used to connect to the grid (which obviously includes older versions, unreleased versions, self-compiled versions, etc.)
- Singularity is now the second most popular viewer in terms of number of user minutes – and there are 165 different versions being used to connect to the grid
- The official viewer is the third most popular viewer in terms of number of user minutes
- Phoenix is the fourth most popular viewer, but with numbers continuing to decline.
Pamela Galli has reported issues with animations, commenting:
I don’t know where else to report this really, so I will just go on record here as saying that I just had to redo the animations in an oven — animations I have used for several years — because they suddenly became borked, and not just in my store, but in other locations. By borked, I mean that the legs started doing random things each time someone “sat” on the oven — crouching one time, leg out to the side another, legs entwined another.
The matter has been raised as JIRA (BUG-204), however, Pamela’s comment caused Monty Linden to reply:
There is work underway on animations and avatars so if anything anomalous is happening, it’s in your best interest to report a bug via JIRA. Gather up as much supporting information as you can and follow the outline here: http://wiki.secondlife.com/wiki/Bug_Tracker.
So if you do encounter peculiar (and reproducible) animation issues, be sure to raise a report.
Performance and Mesh
Concerns are being raised at the level of performance being experienced at Fantasy Faire 2013 due to the time being taken for textures and objects to render – simply because of the load being placed on viewer / server connections during downloads. Part of the problem here is that with services now being switched over to HTTP, those services are coming under increased stress, as Monty Linden commented on the server forum thread:
I visited the host that’s currently running the ‘Crimson Fields’, ‘Fairelands Junction’, and ‘Magnificat’ regions. It was undergoing the heaviest sustained mesh download I’ve looked at as well as a healthy load of textures. The result is a significant delay getting into some HTTP services, something I’ve seen since working on SVC-6760. Some causes are known to me, some may not be revealed. But good bug reports will help identify them.
That mesh “shotguns” the network badly is actually known to Monty – he’s commented on the situation himself numerous times, and it is on his list of HTTP-related work. Events such as Fantasy Faire appear to be exacerbating the issue not only because of the high use of mesh and textures in-world, but also because of the use of things like mesh avatars and attachments. Part of this issue comes down to social education, in much the same was as it is with discouraging people from attending events with script-heavy avatars. Even so, there is still work to be done on the server end of things, and Monty’s ongoing HTTP work will hopefully assist with matters.